Background
A global payments technology leader sought to modernize how merchant statements were analyzed and converted into competitive pricing proposals. Traditionally, statement reviews relied on manual processes that were slow, error-prone, and costly, creating delays in sales cycles and inconsistent outputs. To enhance efficiency, accuracy, and competitiveness, the company partnered with Narwal to build an AI-powered Intelligent Document Processing (IDP) solution leveraging AWS-native tools and modern AI frameworks.
Challenges
The organization encountered significant hurdles with the manual review process:
- Time-Consuming Reviews: Merchant statement analysis was slow and resource-intensive, limiting sales agility.
- Inconsistent Outputs: Lack of standardization across diverse statement formats led to inconsistent fee evaluations.
- Complex Pricing Logic: Legacy Excel-based macros created bottlenecks and introduced errors.
- Low Scalability: Onboarding new statement formats required significant manual effort.
- Accuracy Risks: Manual data entry left room for miscalculations and downstream proposal errors.
Solution
Narwal designed and implemented an AI-Driven Merchant Statement Evaluation Path using the AWS AI stack:
- Automated Data Extraction
- Merchant PDFs ingested into Amazon S3, processed through Textract OCR and Bedrock LLMs (Claude/Titan), extracting 42+ key attributes including transaction volumes, rates, and fees.
- Standardized Fee Taxonomy
- Developed a universal taxonomy covering 20+ fee categories (e.g., transaction charges, compliance, downgrades, closure fees) in collaboration with subject matter experts, enabling consistent comparisons.
- Pricing Engine Modernization
- Converted legacy Excel macros into Python APIs for accurate effective rate calculations, scenario simulations, and structured outputs in JSON and optional PDF/DOCX formats.
- Seamless Workflow Integration
- AI outputs integrated directly with the company’s pricing calculator, streamlining proposal generation.
- Scalable, Secure Architecture
- Built on AWS tools (Textract, Bedrock, SageMaker, Comprehend), vector databases, and Langchain/LangGraph frameworks, ensuring scalability, accuracy, and security.
Outcomes
The initiative delivered clear performance improvements:
- High Accuracy: Achieved 95%+ extraction precision across pilot merchants.
- Faster Turnaround: Reduced processing time to under 30 seconds per statement.
- Proposal Automation: Enabled 90%+ of proposals to be generated without manual edits.
- Scalable Onboarding: New merchant statement formats could be onboarded in less than one week.
- Operational Efficiency: Reduced reliance on manual reviews while enhancing consistency and competitiveness.
- Sales Impact: Accelerated pricing cycles positioned the company to improve win rates in highly competitive markets.
Conclusion
Narwal’s AI-powered IDP solution transformed merchant statement analysis from a manual, error-prone process into a fast, accurate, and scalable workflow. By combining AWS-native AI services, advanced fee taxonomy, and Pythonized pricing logic, the client is now equipped with a differentiated, AI-enabled tool that accelerates sales cycles, improves competitiveness, and builds long-term operational scalability.
